Some mathematical pitfalls in robotics

Robotics research is exciting and frustrating at the same time.
Sometimes, it appears easy to mimic but at the same time is very difficult to generalize or reproduce. This talk tries to shed some light into this topic from a mathematical point of view by highlighting some common pitfalls prevalent in robotics education and research. These cover problems in the area of statistics/ML for robotics, whole body control, hybrid motion-force control, dynamic parameter identification etc. To address such pitfalls, a framework based on some philosophical ideas from Sir Roger Penrose and David Deutsch is presented which can lead to better robotics research. This framework promotes quest for invariance in robotics and it is further demonstrated how search for such invariance properties can lead to solution to some of the aforementioned problems.